DESCRIPTION: (Applicant's Abstract) NIDA and other agencies have recently funded numerous longitudinal observational cohort studies of health/behavior in illicit drug users that have already provided invaluable information. However, new analytic methods for this observational cohort data are needed to answer important questions about (i) the current/changing patterns of morbidity and death in drug users and (ii) the usage, costs, and benefits of interventions. As observational cohort data have not existed in large amounts until recently, methods of analysis for such data are not well developed. This project proposes to develop needed statistical/epidemiological methods to further analyze existing observational drug user cohort data. A comprehensive statistical/epidemiological team will develop and apply methodology with support from drug use study experts. Ten large studies will collaborate in modeling and applications towards their data. Four major areas of statistical/epidemiological research relevant to longitudinal observational drug user data will be developed: (1) Description and analysis of competing events that constitute the natural history of drug use related disease and morbidity (i.e., AIDS, tuberculosis, hepatitis, overdoses, endocarditis, etc.); (2) Adjustment for epidemiological biases to allow comparison of HIV disease progression in drug users versus other risk groups and estimation of treatment effects on this disease progression; (3) Nonparametric models to more accurately describe complicated relationships (linear, survival and other) occurring in observational drug user data; (4) Analysis of repeated events such as costs, duration and timing of repeating drug related conditions (i.e., hospitalizations, sexually transmitted diseases, relapses to addiction). This project will develop epidemiological/statistical tools to analyze illicit drug user data and document these methods for future use (i.e., with software). It will perform simultaneous comparative analysis of research questions in many observational drug cohort datasets. These analyses will resolve key medical and scientific issues concerning the epidemiology of HIV-1 and other diseases in drug users.